Deep Learning Defect Detection in Display Panel Stacked Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional defect detection methods in display panel manufacturing are limited by human error and data loss, and they struggle to provide detailed location information of defects across multiple layers.
Innovation Solution
A deep learning-based method and system that utilize TEM, SEM, STEM, SIM, and EDS images to learn defect and layer information, enabling automatic detection of defect locations across multiple layers in a stacked structure of a display panel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional manual defect detection methods are used, then engineers can identify defects, but data information is lost and determination time increases
Solution Approach 1:
The patent replaces manual mechanical inspection by engineers with an automated deep learning-based image processing system. The system uses neural networks to automatically analyze microscope images and detect defects, eliminating the time-consuming manual review process while maintaining or improving detection accuracy through consistent algorithmic application.
Solution Approach 2:
The patent creates digital copies of defect images and inspection data that can be stored, analyzed, and reviewed without losing information. By digitizing the inspection process and storing image data with detected defect information, the system prevents data loss and enables repeated analysis without additional time cost.
2Ease of operation
If conventional object detection technology is used, then simple object type and location can be checked, but detailed location information in single image is limited
Solution Approach 1:
The patent transitions from two-dimensional single-image analysis to three-dimensional multi-layer spatial analysis. By processing images from multiple microscope types (TEM, SEM, STEM, SIM) that capture different depths and perspectives of the stacked structure, the system extracts detailed 3D location information about defects, including which specific layer and precise position within the layered structure.
Solution Approach 2:
The patent segments the inspection task across multiple specialized image sources, each providing different information about the stacked structure. TEM images provide lattice-level detail, SEM images provide surface morphology, STEM images provide compositional information, and SIM images provide secondary ion data. The deep learning system integrates these segmented information sources to achieve comprehensive defect localization.
3Reliability
If multiple microscope images are analyzed manually, then comprehensive defect information can be obtained, but human error increases and cost increases
Solution Approach 1:
The patent merges multiple image processing functions and data sources into a unified deep learning system. The neural network simultaneously processes images from TEM, SEM, STEM, and SIM microscopes, integrating the information to detect and locate defects across the stacked structure. This consolidation reduces human error by eliminating manual interpretation while managing complexity through automated integration.
Solution Approach 2:
The deep learning system performs self-service by automatically analyzing all microscope images, detecting defects, and determining their locations without human intervention. The system trains on labeled data and then autonomously applies the learned patterns to new inspection data, reducing both human error and the operational complexity of coordinating manual analysis across multiple instruments.
Data Source
AI summary
A method of detecting a defect in a stacked structure of a display panel includes collecting a first image of the defect and a plurality of layers in the stacked structure from a database, learning a defect information of the defect and a layer information of the layers using a deep learning model based on the first image and detecting a location of the defect among the layers by the defect information and the layer information.